M. J. Carey, L. M. Haas, and M. Livny. Tapes hold data, too: Challenges of tuples on tertiary store. In Proc. ACM SIGMOD Int. Conf. on Management of Data, pages 413--7, Washington, DC, 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....storage. Although tertiary storage, in particular magnetic tape, has historically been used primarily for archival or backup purposes, the exploding storage requirements and the high cost of secondary storage are forcing computer architects and designers to re evaluate the role of tertiary storage[CHL93] Work supported by a research grant from NSF ARPA NASA IRI9411330 and NSF instrumentation grant CDA9421978. 1 Commercially available automatic tape and disk libraries or jukeboxes (e.g. EXA96, SON96, Qua96] provide automated access to large amounts of tertiary storage. These libraries can ....

M. J. Carey, L. M. Haas, and M. Livny. Tapes hold data, too: Challenges of tuples on tertiary store. In Proc. ACM SIGMOD Int. Conf. on Management of Data, pages 413--7, Washington, DC, 1993.

....one, is of growing importance. Examples include optical disk jukeboxes and automated magnetic tape libraries. New media such as optical tapes will continue to turn up over time. Nowadays, tertiary storage can hardly be met in DBMSs. Note that we think of tertiary devices as first class citizens [1], i.e. storage under full control of a DBMS, and not just for backup purposes. One reason for neglecting it is that DBMSs insist on random block access, which performs poorly on sequential devices such as tapes. The significant latency of tertiary storage necessitates larger I O granulates. ....

....to be integrated in the DBMS. Managing WORM media requires completely different algorithms for (simulated) updating, clustering, and indexing. Nevertheless, the need for long term storage of massive data and attractive cost per megabyte push the research on tertiary storage support also for DBMSs [1]. We see two basic questions here: How to integrate tertiary storage media into the DBMS and how to let the user take advantage of them. We favor that the first question be answered by DBMS extensibility as studied in STARBURST [10] or especially in the device extensible version of POSTGRES [18] ....

M. J. Carey, L. M. Haas, and M. Livny. Tapes hold data, too: Challenges of tuples on tertiary store. In ACM SIGMOD, pages 413--417, Washington, 1993.

....storage devices such as automatic tape libraries and optical disk jukeboxes. These devices introduce media loading delays of several seconds or more and tape drives involve very long search times. Integrating archival tertiary storage with regular storage is a fairly well explored research topic [3]. However, supporting guaranteed realtime access in such an environment remains a difficult research problem. Christodoulakis, et al. describe the Object Virtual Machine (OVM) which manages a three level multimedia store consisting of main memory, magnetic disks and tertiary storage (i.e. ....

Carey, M.J., Haas, L.M. and Livny, M.: Tapes Hold Data, Too: Challenges of Tuples on Tertiary Store. 1993 ACM SIGMOD Intl. Conf. on Management of Data, pp. 413-417.

....medium are stream oriented. Fast sequential access to data is crucial; transactional updates, fast access to random records, and concurrency control are not. A highly tuned C program can outperform a general purpose RDBMS on this workload. ffl RDBMSs do not usually handle data on tape well [3]. Non clustered indices will not work for traffic data. Worse, traffic analysis data is often used only a few times (or once) so load time is a significant cost. Finally, network traffic traces contain many small records with fields a few bits wide, so per tuple overheads can noticeably increase ....

....queries, Tribeca performed even better. The small cost is far outweighed by the flexibility and convenience of changing small simple queries rather than re writing C code to perform different analyses. 4 Related Work The difficulties in using relational databases stored on tape are overviewed in [3]. Sarawagi [11] modifies a relational query optimizer to consider large tape archives in its cost formula and caches tape data on faster storage. Video on demand systems [4] might use tape storage, but in these workloads many users randomly access independent large objects instead of sequences of ....

M. Carey, L. Haas, and M. Livny. Tapes hold data, too: Challenges of tuples on tertiary store. In Proc. ACM SIGMOD Conference, 1993.

....support for many users making small transactional database updates 1 . Maintaining great volumes within a DBMS is difficult. A number of researchers are attempting to address this problem [SSU91, BCV93] Some are investigating approaches including employing tertiary storage within a DBMS [Sto94, CHL93, Moh93] 2 , while others are examining parallel database systems [DG92] Successful commercial parallel relational database systems have been deployed [BFG 95, Pag92] According to Gesman [Ges96] the exploitation of parallelism within OODBMS has been almost completely ignored (p. 852) ....

Michael J. Carey, Laura M. Haas, , and Miron Livny. Tapes Hold Data, Too: Challenges of Tuples on Tertiary Store. In Proceedings of the 1993 ACM SIG- 157 MOD International Conference on Management of Data, volume 22, Washington, DC., May 1993.

.... Although tertiary storage, in particular magnetic tape, has historically been used primarily for archival or backup purposes, the exploding storage requirements and the higher cost of secondary storage are forcing computer architects and designers to re evaluate the role of tertiary storage[CHL93] Commercially available automatic tape and disk libraries or juke boxes (e.g. EXA96, SON96, Qua96] provide automated access to large amounts of tertiary storage. These libraries can hold hundreds or thousands of media. Throughout this paper we will use the term medium to refer to a tertiary ....

M. J. Carey, L. M. Haas, and M. Livny. Tapes hold data, too: Challenges of tuples on tertiary store. In Proc. ACM SIGMOD Int. Conf. on Management of Data, pages 413--7, Washington, DC, 1993.

....demonstrate that for most source segments x, there are about 300 destination segments y such that locate time(x,y 1) exceeds locate time(x,y) by 25 seconds. It is recognized in the literature that tape storage is of vital importance. Several authors discuss the need for progress in tape storage [9, 18, 2]. A reference model for hierarchical storage appears in [3] and [10, 6] discuss the role of hierarchical storage management in video on demand systems. 11] presents a logstructured filesystem that spans disk and tape storage. Striped tape organizations are studied by [4, 7] 14] describes join ....

Carey, M. J., Haas, L. M., and Livny, M. Tapes hold data, too: challenges of tuples on tertiary store. In Proceedings SIGMOD International Conference on Management of Data (Washington, DC, May 26--28 1993), pp. 413--417.

....these are stable logged and maintained in such a way that query results are based on the most recent state of the database. ffl Since the time to mount a tape and seek data within a tape is in the order of few tens of seconds, access to tapes must be even better optimized than to disks [SSU90, CHL93] We develop novel approaches to scheduling the tape access requests of dynamically arriving concurrent queries such that the average response time of queries is minimized. Our approaches minimize the number of tape mounts by reading writing data from to a mounted tape opportunistically. Which ....

M. Carey, L. Haas, and M. Livny. Tapes Hold Data, Too: Challenges of Tuples on Tertiary Store. In Proc. of SIGMOD Intl. Conference on Management of Data, pages 413--419, 1993.

....for tertiary memory, a lot of issues related to tertiary memory specific performance optimization still remain unexplored. Inclusion of tertiary memory devices in the storage hierarchy requires a rethinking of many design decisions made for a conventional database system. Many database researchers [1, 11, 5, 8] have reached consensus regarding the need of a database system specially optimized for manipulating data stored on a tertiary memory device. In this paper, we will see how we modified the design of an existing Filesystem Disk cache User Tertiary Memory (a) Only Filesytem Filesystem (c) Only ....

M.J. Carey, L.M. Haas, and M. Livny. Tapes hold data, too: challenges of tuples on tertiary store. SIGMOD Record, 22(2):413--417, 1993.

....1 Introduction Applications manipulating large volumes of data are growing in number: earth observation systems, historical data base systems, statistical data collections and image and video storage systems are a few examples. There is increasing consensus amongst database researchers [Sto91] CHL93] Sel93] Moh93] regarding the need of a database controlled tertiary memory for storing massive amounts of data. A major limitation of traditional DBMSs is the assumption that all data resides on magnetic disk or main memory. Therefore all optimization decisions are oriented towards this ....

M.J. Carey, L.M. Haas, and M. Livny. Tapes hold data, too: challenges of tuples on tertiary store. SIGMOD Record, 22(2):413--417, 1993.

....towards this technology. Tertiary memory, if used at all, functioned only as an archival storage system to be written once and rarely read. With the inclusion of tertiary memory as an active part of the storage hierarchy it is necessary to rethink the optimization decisions made by a DBMS [STO91a] [CAR93]. In this paper, we propose improvements to existing query execution strategies to adapt them to tertiary storage devices. Tertiary memory devices have very different performance characteristics than magnetic disks. A typical device consists of a large number of storage media (tapes or disk ....

M.J. Carey, L.M. Haas, and M. Livny. Tapes hold data, too: challenges of tuples on tertiary store. SIGMOD Record, 22(2):413--417, 1993.

.... of data cannot be stored cost effectively on magnetic disks [SD91, SSU95] In view of applications like EOSDIS and other applications like data warehouses [Ome92] image [RFJ 93, OS95] and video storage systems [FR94] there is increasing consensus among database researchers [SSU95, Sto91a, CHL93, Sel93, Moh93] for supporting tertiary memory devices [Ran91] Not only are all these applications huge, they also require efficient querying and data management facilities, making it necessary to deploy database systems instead of relying on conventional file oriented mass storage systems [N ....

....concentrate our discussion on the use of tertiary memory devices with a DBMS. As a result of the increasing demand for handling larger and larger datasets, the database community started realizing the need for handling tertiary memory devices and many researchers proposed doing so [SSU95, Sto91a, CHL93, Sel93, Moh93] As a result some DBMSs started providing support for storing and accessing data on tertiary memory devices. Such DBMSs can be classified into two categories. The first and the more common category consists of DBMSs that do not exercise direct control of the tertiary memory device ....

M.J. Carey, L.M. Haas, and M. Livny. Tapes hold data, too: challenges of tuples on tertiary store. Proc. ACM SIGMOD International Conference on Management of Data, 22(2):413--417, 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC